Presentation | 2007/7/19 Acoutstic Modeling Based on Model Structure Annealing for Speech Recognition Sayaka SHIOTA, Kei HASHIMOTO, Heiga ZEN, Yoshihiko NANKAKU, Akinobu LEE, Keiichi TOKUDA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | This paper proposes a joint optimization technique of phonetic decision trees and state sequences for HMM-based speech recognition. In context-dependent models (i.e., triphone HMMs), the decision tree based context clustering is applied to extract an optimal parameter tying structure given HMM state sequences. On the other hand, the DAEM(Deterministic Annealing Expectation Maximization) algorithm has been proposed to estimate optimal state sequences in the training of HMMs. However, these techniques optimize phonetic decision trees and HMM state sequences independently with keeping the other fixed. To overcome these problems, we propose model structure annealing in which the DAEM algorithm is applied to optimize a probabilistic model including the multiple decision trees as a hidden variable. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Continuous speech recognition / Acoustic modeling / Context clustering / Phonetic decision tree / Deterministic annealing |
Paper # | SP2007-35 |
Date of Issue |
Conference Information | |
Committee | SP |
---|---|
Conference Date | 2007/7/19(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Speech (SP) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Acoutstic Modeling Based on Model Structure Annealing for Speech Recognition |
Sub Title (in English) | |
Keyword(1) | Continuous speech recognition |
Keyword(2) | Acoustic modeling |
Keyword(3) | Context clustering |
Keyword(4) | Phonetic decision tree |
Keyword(5) | Deterministic annealing |
1st Author's Name | Sayaka SHIOTA |
1st Author's Affiliation | Depertment of Computer Science and Engineering, Nagoya Institute of Technology() |
2nd Author's Name | Kei HASHIMOTO |
2nd Author's Affiliation | Depertment of Computer Science and Engineering, Nagoya Institute of Technology |
3rd Author's Name | Heiga ZEN |
3rd Author's Affiliation | Depertment of Computer Science and Engineering, Nagoya Institute of Technology |
4th Author's Name | Yoshihiko NANKAKU |
4th Author's Affiliation | Depertment of Computer Science and Engineering, Nagoya Institute of Technology |
5th Author's Name | Akinobu LEE |
5th Author's Affiliation | Depertment of Computer Science and Engineering, Nagoya Institute of Technology |
6th Author's Name | Keiichi TOKUDA |
6th Author's Affiliation | Depertment of Computer Science and Engineering, Nagoya Institute of Technology |
Date | 2007/7/19 |
Paper # | SP2007-35 |
Volume (vol) | vol.107 |
Number (no) | 165 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |